🚀 go-pugleaf

On 8 Jan 91 17:33:02 GMT, anw@maths.nott.ac.uk (Dr A. N. Walker) said:

	[ ... on whether invalid declarations can be forbidden
	on purely syntactive grounds, which it cannot be done
	with a context free grammar ... ]

anw> As others have pointed out, this is easy with a two-level grammar.
anw> For details, see the Revised Report, or [better!]  "Grammars of
anw> Programming Languages" by Cleaveland and Uzgalis (Elsevier, 1977).

Ah, here you take out the book which is the nuclear weapon of context
sensitive grammar debates. I agree with what you say, as far as it goes,
but you omit to remind us that that book clearly demonstrates what the
*semantics* of a language can easily be formalized with a two level
grammar, which throws a spanner in the well oiled structure of this type
of debate (Bernstein writes provocative, edgy but sensible things,
everybody else stop chewing their sandwich and jumps on him with glee
:->).

Context sensitive grammar generators (they do exist!) can do pretty
impressive things -- just consider that people have been using simple
context free grammars to do, in favourable conditions, code generation!

Two level or attribute grammars are strange beasts indeed, at least from
the usual CS perspective under which grammars stop at context free.

End of debate while a lot of people stop jumping on Bernstein and rush
to the library (not the bookshop -- out of print for a long time) to get
a copy of the book over which to finish munching their sandwich? :-)
--
Piercarlo Grandi                   | ARPA: pcg%uk.ac.aber.cs@nsfnet-relay.ac.uk
Dept of CS, UCW Aberystwyth        | UUCP: ...!mcsun!ukc!aber-cs!pcg
Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk